The act of composing is best described as a set of distinguishable processes that interact. There are four methods for researching these processes: (1) behavior protocols, in which subjects are observed but are not asked to report their thought processes verbally; (2) directed reports, in which subjects are asked to explain how they performed a task; (3) directed reports, in which the subject is asked specific questions while performing a task; and (4) thinking aloud protocols, in which subjects report on anything they are thinking while performing a task. These process tracing methods are better than input/output methods because they can pinpoint difficulties when the writer encounters them, and they offer valuable opportunities for scientific exploration. In addition, there are aspects of writing that are difficult to observe without these methods. Three major objections have been raised about the use of verbal reports as data: verbal reports are unreliable because people are not conscious of their cognitive processes, reporting them verbally distorts them, and verbal reports are incomplete and not objective. It is interesting that verbal protocols are singled out for criticism on the grounds of incompleteness, since they are characteristically more complete than most other methods to which they are compared. (HTH)